Publication number | US6347389 B1 |

Publication type | Grant |

Application number | US 09/274,406 |

Publication date | Feb 12, 2002 |

Filing date | Mar 23, 1999 |

Priority date | Mar 23, 1999 |

Fee status | Paid |

Also published as | WO2000057561A1 |

Publication number | 09274406, 274406, US 6347389 B1, US 6347389B1, US-B1-6347389, US6347389 B1, US6347389B1 |

Inventors | Keith G. Boyer |

Original Assignee | Storage Technology Corporation |

Export Citation | BiBTeX, EndNote, RefMan |

Patent Citations (8), Non-Patent Citations (1), Referenced by (36), Classifications (8), Legal Events (4) | |

External Links: USPTO, USPTO Assignment, Espacenet | |

US 6347389 B1

Abstract

The pipelined high speed Reed-Solomon error/erasure decoder processes multiple code words in a pipelined fashion. The pipelined high speed Reed-Solomon error/erasure decoder is designed to process Reed-Solomon encoded words that have been corrupted in a digital system by processing errors as well as erasures through a simple iterative modified syndrome process. The iterative nature of this method provides for limited computational effort at each step of the pipeline. This allows the pipelined high speed Reed-Solomon error/erasure decoder to easily handle full or shortened Reed-Solomon codes, as well as parallel processing to achieve higher data rates. The iterative modified syndrome process is one of the pipelined steps. It relieves the erasure pre-shifting burden from the Berlekamp/Massey synthesis process, which reduces the number of cycles required at that stage of processing. The decoder proceeds classically with a Chien Search for any remaining error locations. This approach allows the pipelined high speed Reed-Solomon error/erasure decoder to relay early information on all error locations. The final stage of the decoding process is a parallel, iterative solution to Forney's equation for the calculation of the error magnitudes.

Claims(28)

1. A system for detecting and correcting both erasures and errors that are contained in a data transmission that includes multiple error correcting codes to identify said both erasures and errors, comprising:

means for storing said data transmission;

means, responsive to said data transmission, for generating syndrome data;

means, responsive to said data transmission and said generated syndrome data, for generating location data representative of error and erasure locations in said data transmission;

means, responsive to said generated syndrome data and said location data, for creating an associated error and erasure magnitude for each respective error and erasure location in said data transmission; and

means, responsive to said location data, said error and erasure magnitudes, and said data transmission, for correcting each said both erasures and errors contained in said data transmission.

2. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 1 , further comprising:

means for producing a digital indication that a code word symbol of said data transmission being received has been determined to contain an erasure; and

means, responsive to said digital indication, for identifying locations of said erasure in said code word symbol.

3. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 1 , wherein said means for generating location data comprises:

means for creating an error locator polynomial that represents said errors contained in said data transmission;

means for determining roots of said error locator polynomial representing said errors; and

means for creating data indicative of error locations of said error in said data transmission.

4. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 3 wherein said means for creating an error locator polynomial comprises:

means for executing a Berlekamp/Massey process to create said error locator polynomial that represents said errors contained in said data transmission.

5. The system for detecting and correcting both erasures and errors that are contained in a date transmission of claim 3 wherein said means for determining roots comprises:

means for executing a Chien Search process to determine said roots of said error locator polynomial.

6. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 3 wherein said means for creating data indicative of error locations comprises:

means for executing a Sigma process on said roots of said error locator polynomial.

7. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 1 , wherein said means for correcting comprises:

means for reading code words of said data transmission from said means for storing; and

means for correcting each said both erasures and errors contained in said code words in said data transmission using said error and erasure location and a corresponding said error and erasure magnitude.

8. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 1 wherein said means for generating syndrome data comprises:

means for generating a set of syndrome data indicative of at least one erasure contained in said data transmission; and

means for modifying said set of syndrome data to generate modified syndrome data indicative of at least one error contained in said data transmission.

9. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 8 wherein said means for generating syndrome data further comprises:

means for iteratively activating said means for generating a set of syndrome data and said means for modifying said set of syndrome data to correct a plurality of erasures and errors contained in said data transmission.

10. A method for detecting and correcting both erasures and errors that are contained in a data transmission that includes multiple error correcting codes to identify said both erasures and errors, comprising the steps of:

storing said data transmission in a memory;

generating, in response to said data transmission, syndrome data;

generating, in response to said data transmission and said generated syndrome data, location data representative of error and erasure locations in said data transmission;

creating, in response to said generated syndrome data and said location data, an associated error and erasure magnitude for each respective error and erasure location in said data transmission; and

correcting, in response to said location data, said error and erasure magnitudes, and said data transmission, each said both erasures and errors contained in said data transmission.

11. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 10 , further comprising the steps of:

producing a digital indication that a code word symbol of said data transmission being received has been determined to contain an erasure; and

identifying locations of said erasure in said code word symbol.

12. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 10 , wherein said step of generating location data comprises:

creating an error locator polynomial that represents said errors contained in said data transmission;

determining roots of said error locator polynomial representing said errors; and

creating data indicative of error locations of said error in said data transmission.

13. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 12 wherein said step of creating an error locator polynomial comprises:

executing a Berlekamp/Massey process to create said error locator polynomial that represents said errors contained in said data transmission.

14. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 12 wherein said step of determining roots comprises:

executing a Chien Search process to determine said roots of said error locator polynomial.

15. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 12 wherein said step of creating data indicative of error locations comprises:

executing a Sigma process on said roots of said error locator polynomial.

16. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 10 , wherein said step of correcting comprises:

reading code words of said data transmission from said means for storing; and

correcting each said both erasures and errors contained in said code words in said data transmission using said error and erasure location and a corresponding said error and erasure magnitude.

17. The method of detecting and correcting both erasures and errors that are contained in a data transmission of claim 10 wherein said step of generating syndrome data comprises:

generating a set of syndrome data indicative of at least one erasure contained in said data transmission; and

modifying said set of syndrome data to generate modified syndrome data indicative of at least one error contained in said data transmission.

18. The method of detecting and correcting both erasures and errors that are contained in a data transmission of claim 17 wherein said step of generating syndrome data further comprises:

iteratively activating said step of generating a set of syndrome data and said step of modifying said set of syndrome data to correct a plurality of erasures and errors contained in said data transmission.

19. A system for detecting and correcting both erasures and errors that are contained in a data transmission that includes multiple error correcting codes to identify said both erasures and errors, comprising:

means for storing said data transmission;

means, responsive to said data transmission, for producing a digital indication that a code word symbol of said data transmission being received has been determined to contain an erasure;

means, responsive to said digital indication, for identifying locations of said erasure in said code word symbol;

means for generating data representative of error locations in said data transmission;

means for creating an associated error and erasure magnitude for each respective error and erasure location in said data transmission; and

means, responsive to said error and erasure locations, said error and erasure magnitudes, and said data transmission, for correcting each said both erasures and errors contained in said data transmission, comprising:

means for reading code words of said data transmission from said means for storing; and

means for correcting each said both erasures and errors contained in said code words in said data transmission using said error and erasure location and a corresponding said error and erasure magnitude.

20. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 19 , wherein said means for generating data comprises:

means for creating ah error locator polynomial that represents said errors contained in said data transmission;

means for determining roots of said error locator polynomial representing said errors; and

means for creating data indicative of error locations of said error in said data transmission.

21. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 20 wherein said means for creating an error locator polynomial comprises:

means for executing a Berlekamp/Massey process to create said error locator polynomial that represents said errors contained in said data transmission.

22. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 20 wherein said means for determining roots comprises:

means for executing a Chien Search process to determine said roots of said error locator polynomial.

23. The system for detecting and correcting both erasures and errors that are contained in a data transmission of claim 20 wherein said means for creating data indicative of error locations comprises;

means for executing a Sigma process on said roots of said error locator polynomial.

24. A method for detecting and correcting both erasures and errors that are contained in a data transmission that includes multiple error correcting codes to identify said both erasures and errors, comprising the step of:

storing said data transmission in a memory;

producing, in response to said data transmission, a digital indication that the code word symbol of said data transmission being received has been determined to contain an erasure;

identifying locations of said erasure in said code word symbol;

generating data representative of error locations in said data transmission;

creating an associated error and erasure magnitude for each respective error and erasure location in said data transmission; and

correcting, in response to said error and erasure locations, said error and erasure magnitudes, and said data transmission, each said both erasures and errors contained in said data transmission, comprising:

reading code words of said data transmission from said means for storing; and

correcting each said both erasures and errors contained in said code words in said data transmission using said error and erasure location and a corresponding said error and erasure magnitude.

25. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 24 , wherein said step of generating data comprises:

creating an error locator polynomial that represents said errors contained in said data transmission;

determining roots of said error locator polynomial representing said errors; and

creating data indicative of error locations of said error in said data transmission.

26. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 25 wherein said step of creating an error locator polynomial comprises:

executing a Berlekamp/Massey process to create said error locator polynomial that represents said errors contained in said data transmission.

27. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 25 wherein said step of determining roots comprises:

executing a Chien Search process to determine said roots of said error locator polynomial.

28. The method for detecting and correcting both erasures and errors that are contained in a data transmission of claim 25 wherein said step of creating data indicative of error locations comprises:

executing a Sigma process on said roots of said error locator polynomial.

Description

This invention relates to a system that decodes a class of multiple error correcting codes known as Reed-Solomon Codes, which are typically used within digital communication systems to detect corrupted data contained in the digital transmission and correct these corrupted data. As part of the Reed-Solomon error correcting process, redundancy is added to the transmitted data, such that upon receipt of a data transmission, the possibly corrupted data and redundancy are mathematically processed to correct the corrupted data in the data transmission.

It is a problem in high speed decoding of multiple error correcting codes, known as Reed-Solomon codes, with or without erasure pointers, to reduce the number of computational steps that are required at each stage of decoding. It is, likewise, a problem to reduce the number of combinatorial operations that are required to exchange data between any two hardware registers that are operational within the Reed-Solomon decoder.

Reed-Solomon Codes are used within digital communication systems, such as fiber optic transmission systems, optical and magnetic data storage systems, where the corruption of data being transferred between any two points affects system reliability. As part of the Reed-Solomon error correcting process, redundancy is added to the transmitted data, such that upon receipt of a data transmission the possibly corrupt data and redundancy are mathematically processed to correct errors in the data transmission. Prior art Reed-Solomon decoders use both Berlekamp/Massey and Chien Search processes to detect and correct the errors that are present in the data transmission. These prior art Reed-Solomon decoders are primarily concerned with the total number of computations required to perform the data transmission decoding process, and the amount of hardware required to perform the data transmission decoding process. The Berlekamp/Massey synthesis process includes an erasure pre-shifting burden which adds to the processing complexity. These prior art Reed-Solomon decoders typically lack explicit early indication of error locations and do not provide early erasure information from the Berlekamp/Massey and Chien Search processes.

The above described problems are solved and a technical advance achieved by the pipelined high speed Reed-Solomon error/erasure decoder of the present invention which processes multiple code words in a pipelined fashion. The piplelined high speed Reed-Solomon error/erasure decoder processes Reed-Solomon encoded data words to detect the presence of data words that have been corrupted in a digital system and processes corrupted data, that comprises both errors as well as erasures, through a simple iterative modified syndrome process. The iterative nature of this method provides for limited computational effort at each step of the pipeline. This allows the pipelined high speed Reed-Solomon error/erasure decoder to easily handle full or shortened Reed-Solomon codes, as well as parallel processing to achieve higher data rates. The iterative modified syndrome process is one of the pipelined steps. It relieves the erasure pre-shifting burden from the Berlekamp/Massey synthesis process, which reduces the number of cycles required at that stage of processing. The pipelined high speed Reed-Solomon error/erasure decoder then proceeds classically with a Chien Search for any remaining error locations. This approach allows the pipelined high speed Reed-Solomon error/erasure decoder to relay early information on all error locations. The final stage of the decoding process is a parallel, iterative solution to Forney's equation for the calculation of the error magnitudes.

This pipelined high speed Reed-Solomon error/erasure decoder emphasizes the number of iterations (or clock cycles) and the number of combinatorial operations between each hardware register, which is accomplished by combining iterative LFSR solutions for syndrome modification and error magnitude processing to the well known Berlekamp/Massey and Chien Search iterative hardware solutions. Additionally, in augmenting the known erasure locations with a classic Berlekamp/Massey—Chien Search for any remaining error locations, the pipelined high speed Reed-Solomon error/erasure decoder also provides explicit early indication of all error locations. This can be helpful in providing early erasure information for errors that no erasure location was provided.

FIG. 1 illustrates in block diagram form the pipelined high speed Reed-Solomon error/erasure decoder of the present invention;

FIG. 2 illustrates in block diagram form the Error Locator Coefficient Generator (Sigma Processor)

FIG. 3 illustrates in block diagram form the topology of the Modified Syndrome Processor hardware shift register; and

FIG. 4 illustrates in block diagram form the circuit construction of the Errata Value Processor for generating each of the error magnitudes.

In the encoding of an (n, k) Reed-Solomon code, k data symbols are processed for each n−k=r redundancy symbols. The maximum number of symbols in the code word, which are developed by appending the k data symbols to the r redundancy symbols, is determined by the size of the Galois Field that the Reed-Solomon code is based on. This finite field, known as GF (2^{m}), is defined by a primitive polynomial with order n=2^{m}−1. Each redundancy symbol is an element ∝^{i }of the field GF (2^{m}). The number of symbols in the data transmission data word that a code can correct is equal to the number of redundancy symbols (r). The minimum distance d=r+1 of a redundancy code defines the minimum number of symbols for which every possible code word must differ. Each redundancy symbol can be used mathematically to either find the location or the magnitude of a symbol in error. Consequently, it takes two redundancy symbols to both locate and correct a single error in the data transmission.

A Reed-Solomon decoder should handle corrupted data that comprises both erasures and errors to be effective in correcting data transmissions. When the location of corrupted data is know by any means that is external to a decoding process, it is called an erasure (e), and when the location of corrupted data is not know to the decoding process it is referred to as an error (t). Considering the minimum distance (r) of the redundancy code, for (r) symbols to be corrected, the location of each corrupted data byte must be known. The gives a code constraint of (2t+e≦d−1) for an erasure/error correcting Reed-Solomon decoder.

The pipelined high speed Reed-Solomon error/erasure decoder **100** starts by calculating, in a syndrome generator **101**, the partial syndromes of the received data word v(x)=c(x)+e(x), where c(x) is the encoded word made up of k data symbols and r redundancy symbols, and e(x) is the error word. Since c(x) is a multiple of the generator polynomial G(x), it drops out of the calculation of the partial syndromes, and the syndrome calculations become equivalent to evaluating the error polynomial at the roots of the generator polynomial G(x):

In the case where both errors and erasures are considered by the Reed-Solomon decoding process, the syndrome includes the transform for both the (t) errors and (e) erasures:

With the (e) know erasure locations, the pipelined high speed Reed-Solomon error/erasure decoder **100** then seeks to find up to (t) remaining errors. This requires that the transform syndromes S_{j }be modified by a modified syndrome processor **102** to factor out the effect of the (e) erasure locations. The (e) erasure locations are first used to form the coefficients of the associated erasure locator polynomial:

These coefficients are then processed in an iterative fashion to create the remaining modified transform syndromes T_{j}.

The modified transform syndromes T_{j }can then be used in the classic Berlekamp/Massey **103** and Chien Search **104** iterative solutions to again find up to (t) errors that remain in the data transmission. The (t) newly discovered error locations are combined with the original (e) erasure locations to form the entire explicit set of error locations {X_{j}}. The coefficients of the final error locator polynomial for this set of error locations are then constructed as before, except that the polynomial is created for (t+e) erasure and error locations.

The coefficients of σ(X) are related to the erasure and error locations by the elementary symmetric functions σ_{ji}. Together with the original syndromes, Forney's equation for the error magnitudes {E_{j}} can be solved:

Many decoding circuits for Reed-Solomon codes consider erasure pointers, but as disclosed in U.S. Pat. No. 5,379,305 (Lih-Jyh Weng), the decoder modifies the original transform syndromes for (p+e) erasures according to assumed precalculated coefficients. The pipelined high speed Reed-Solomon error/erasure decoder **100** provides a simple recursive hardware solution to this problem. It is based on an iterative solution to the elementary symmetric functions related to σ(X). This solution shifts recursively for each known erasure location, according to the recursion:

_{i}(*x*)=σ_{i−1} *+X* _{i}σ_{i−1},

where σ_{0}(x)=1.

The recursion shifts to the highest possible order, shifting with zero for any missing erasure locations. In this way, the circuit always has the least significant coefficient (σ_{0}=1) in the same significant position, regardless of the number of erasure locations presented. Additionally, the pipelined high speed Reed-Solomon error/erasure decoder **100** provides a parallel hardware solution that shifts the original syndromes in a barrel shift fashion for each modified syndrome, according to the aforementioned equation for T_{j}.

Similarly, the process of determining the error magnitudes requires the coefficients of the combined error locator polynomial σ(X). In the iterative solution of the above-noted U.S. Pat. No. 5,379,305, these coefficients are again needed. The pipelined high speed Reed-Solomon error/erasure decoder **100**, again, solves the above recursion for σ_{i}(X). Then as a natural extension of this recursion, applies a similar hardware shift register approach to the coefficients required in the numerator and denominator of Forney's solution for the error magnitudes {E_{j}}:

_{ji}=σ_{i} *+X* _{j}σ_{j(i−1)}

The coefficients are retained in a shift register during the generation of the numerator, and used in a similar fashion to generate the denominator. The spirit of the pipelined high speed Reed-Solomon error/erasure decoder **100** differs from previous error evaluators such as that of U.S. Pat. No. 5,430,739 (Shyue-Win Wei, et al.). In this approach a parallel solution to determinants is used to solve the error evaluator problem. This requires fewer clock cycles, but greater combinatorial complexity between each hardware register stage.

FIG. 1 depicts the complete pipelined high speed Reed-Solomon error/erasure decoder **100** with erasure capability. The first block comprises a buffer memory **108** that stores the code words of the received data transmission. These code words are also applied to the Syndrome Generator **101** which generates the transform syndromes according to:

This is a well known process, that yields (2t) syndrome values which are output on data path **113**. The erasure pointers **112** are simply a digital indication that the code word symbol being received has been determined to contain an erasure location. This gives rise to the actual erasure locations **114** being presented to the successive Modified Syndrome process **102**.

The Modified Syndrome process **102** modifies the original (2t) syndromes according to the (e) erasure locations, producing up to (2t) modified syndromes {T_{j}}, depending on the number of erasure locations, which are output on data path **117**. Any unnecessary modified syndromes are padded with zeros as they are passed to the Berlekamp/Massey LFSR process **103**. The Berlekamp/Massey process **103** creates an error locator polynomial that represents the (t) error locations that remain in the data transmission, beyond the (e) known erasure locations. The coefficients of this polynomial are represented by the set {σ_{i}} which is passed on data path **120**. The Berlekamp/Massey process **103** also retains the original syndromes and erasure locations passed to it. These are passed on data paths **118** and **119**, respectively, to the following Chien Search process **104**, where the roots of the error locator polynomial representing the (t) errors are found. The (t) error locations are then concatenated, by the Chien Search process **104**, to the original (e) erasure locations. The result is the entire set of corrupted data locations {X_{j}} which are output on data path **122**. The Berlekamp/Massey process **103** and Chien Search process **104** are not discussed in detail in this embodiment due to fact that they are well documented in the literature.

Together with the original syndromes, the entire set of corrupted data locations {X_{j}} are sent on data paths **121**, **122**, respectively to the Sigma Processor **105**. Here the corrupted data locations are used to create the coefficients of the associated corrupted data locator polynomial {σ_{k}}. These coefficients, together with the original syndromes and the full set of corrupted data locations {X_{j}} are sent on data paths **123**. **124**, **125**, respectively, to the Errata Value Processor **106**, where an associated error magnitude {E_{j}} is created for each respective erasure and error location that corresponds to the corrupted data.

With all the corrupted data locations {X_{j}} and corrupted data magnitudes {E_{j}} defined and output on data paths **126**, **127**, respectively, the original corrupted code word C(x)=c(x)+e(x) is retrieved from buffer memory **108**, used to buffer the associated pipelined delay of the pipelined high speed Reed-Solomon error/erasure decoder **100**, and corrected in each location X_{j }with the value E_{j }by Error Corrector **107**. The result is a corrected version of the corrupted code word C(x) output on data path **128**.

In FIG. 2, the error location “X” register **201** of the corrupted data locator coefficient generator **200** is loaded with the set of error locations {X_{j}}, for either erasures and/or errors. The “σ” register circuit **202** is initialized with “α^{0}=1” in the most significant position, σ_{n}, and zero in all other less significant register positions. The “σ” register circuit **202** is shifted until a “α^{0}=1” appears in the σ_{0 }register position. This may mean that the “σ” register circuit **202** shifts for more cycles than there are available error locations. In this case, the error location register **201** simply shifts zeros into the σ construction. Ultimately, this corrupted data locator coefficient generator **200** is solving the recursion:

_{i}(*X*)=σ_{i −1} *+X* _{i}σ_{i−1},

where σ_{0}(X)=1.

This corrupted data locator coefficient generator **200** is used to provide corrupted data locator coefficients {σ}, for both the Modified Syndrome Processor **102** and the Sigma Processor **105** of FIG. **1**.

Each finite field multiplier is denoted by a circle with a dot in the middle. Likewise, each finite field adder is denoted by a circle with an addition or “+” sign contained within. The addition is simple modulo 2, per bit, while the multiplication is the classic multiplication of two finite field values modulo g(x) (where g(x) is the primitive polynomial which defines the finite field GF (2^{m}).) These mathematical operations are defined the same throughout this document.

FIG. 3 shows the topology of the Modified Syndrome Processor **102** hardware shift register of the pipelined high speed Reed-Solomon error/erasure decoder **100**. In this circuit, the bank of a registers is the set of coefficients generated by the corrupted data locator coefficient generator **200**. These coefficients are based on the set of erasure locations for which the original syndromes are to be modified. The original syndromes are modified according to the following equation:

The register **302** having register positions denoted with “S_{n}” are loaded with the (2t) original syndromes. The order of polynomial in the “σ” register is (e+1), dictated by the number of erasures (e). The circuit shifts for each “j=e+1, e+2, d−1”. For each clock cycle, the original syndrome register is shifted from right to left, and is shown to shift in an end around fashion only so that the original syndromes remain in the register and are not lost. On each cycle, a newly modified syndrome value is shifted into the register **303** having register positions denoted “T_{j}”.

Ultimately, the Modified Syndrome Process **102** is comprised of generating the error locator coefficients associated with (e) erasures, and then using these coefficients to modify the original (2t) syndromes. The modified syndromes can then be used in a Berlekamp/Massey process **103** and Chien Search process **104** to find up to (t) remaining errors.

FIG. 4 shows the circuit construction of the Errata Value Processor **106** for generating each of the error magnitudes {E_{j}}. The same circuit can be used to generate each of the (j=2t) possible error locations, or (2t) of these processors can be constructed so that all (=2t) possible error magnitudes can be calculated in parallel. In the parallel case, the syndrome register denoted “S_{i}” can be shared for all (2t) processors.

The first step in processing the error magnitudes is to generate the coefficients of the error locator polynomial {σ_{k}} associated with the ultimate set of error locations {X_{j}}. These coefficients are, again, calculated using the Error Locator Coefficient Generator circuit of FIG. **2**. These coefficients are then loaded into the σ_{ji }register **401** of FIG. **4**. The syndrome {S} register **402** is loaded with the original (2t) syndromes, and the {X_{j}} register **403** is loaded with the respective error location for the given error magnitude. The (N_{j}) numerator **404** and (D_{j}) denominator **405** registers are loaded with zero, while the X_{j} ^{m }register **406** is loaded with α^{0}=1.

For the first (t) cycles, the σ_{ji }registers shift left, while the σ_{j0 }register position gets the output of “Mux”. The output of “Mux” **407** is the adder output until the last numerator shift (t=num_cycle_n), where the α^{0}=1 result is shifted into the σ_{j0 }register position. Also, for each numerator (t) numerator cycles, the syndrome {S_{i}} register **402** is shifted right one position. This defines the numerator result (N_{j}), and the values σ_{ji}.

For the second (t) cycles, the σ_{ji }registers shift right, and combine with the result of the X_{j} ^{m }register **403**, which shifts together with (D_{j}) for each of the (t) denominator cycles. This defines the denominator result (D_{j}). After the (2t) cycles, the final result {E_{j}}=(Num_{j}/Den_{j}) is accomplished via a combinatorial invert and multiply.

The pipelined high speed Reed-Solomon error/erasure decoder processes multiple code words in a pipelined fashion. The pipelined high speed Reed-Solomon error/erasure decoder processes Reed-Solomon encoded words that have been corrupted in a digital system. The pipelined high speed Reed-Solomon error/erasure decoder processes errors as well as erasures through a simple iterative modified syndrome process. The iterative nature of this processing method provides for limited computational effort at each step of the pipeline. This allows the pipelined high speed Reed-Solomon error/erasure decoder to easily handle full or shortened Reed-Solomon codes, as well as parallel processing to achieve higher data rates.

Patent Citations

Cited Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|

US5323402 * | Feb 14, 1991 | Jun 21, 1994 | The Mitre Corporation | Programmable systolic BCH decoder |

US5379305 | Jul 20, 1992 | Jan 3, 1995 | Digital Equipment Corporation | Error correction system with selectable error correction capabilities |

US5430739 | Jan 22, 1992 | Jul 4, 1995 | National Science Council | Real-time Reed-Solomon decoder |

US5541937 * | Dec 23, 1994 | Jul 30, 1996 | Canon Kabushiki Kaisha | Apparatus for uniformly correcting erasure and error of received word by using a common polynomial |

US5942005 * | Apr 8, 1997 | Aug 24, 1999 | International Business Machines Corporation | Method and means for computationally efficient error and erasure correction in linear cyclic codes |

US6119262 * | Aug 19, 1997 | Sep 12, 2000 | Chuen-Shen Bernard Shung | Method and apparatus for solving key equation polynomials in decoding error correction codes |

US6122766 * | Oct 16, 1997 | Sep 19, 2000 | Matsushita Electric Industrial Co., Ltd. | Reed-Solomon decoder having a three-stage pipeline structure |

US6131178 * | Mar 4, 1998 | Oct 10, 2000 | Mitsubishi Denki Kabushiki Kaisha | Error correcting decoding apparatus of extended Reed-Solomon code, and error correcting apparatus of singly or doubly extended Reed-Solomon codes |

Non-Patent Citations

Reference | ||
---|---|---|

1 | Blaum M. et al., "New Array Codes For Multiple Phased Burst Correction"; IEEE Transactions on Information Theory, US, IEEE Inc. New York, vol. 39, No. 1, 1993, pp. 66-77. |

Referenced by

Citing Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|

US7010739 | Nov 26, 2002 | Mar 7, 2006 | Marvell International Ltd. | Error evaluator for inversionless Berlekamp-Massey algorithm in Reed-Solomon decoders |

US7020826 * | Jan 23, 2002 | Mar 28, 2006 | Thomson Licensing | Intra-decoder component block messaging |

US7051267 | Nov 26, 2002 | May 23, 2006 | Marvell International Ltd. | Efficient high-speed Reed-Solomon decoder |

US7096409 | Jul 30, 2003 | Aug 22, 2006 | Hewlett-Packard Development Company, L.P. | Reed-solomon decoder and decoding method for errors and erasures decoding |

US7222287 * | Feb 26, 2004 | May 22, 2007 | Sony Corporation | Decoder, error position polynomial calculation method, and program |

US7228489 | Dec 26, 2003 | Jun 5, 2007 | Storage Technology Corporation | Soft viterbi Reed-Solomon decoder |

US7228490 | Feb 19, 2004 | Jun 5, 2007 | Quantum Corporation | Error correction decoder using cells with partial syndrome generation |

US7249310 | Oct 4, 2005 | Jul 24, 2007 | Marvell International Ltd. | Error evaluator for inversionless Berlekamp-Massey algorithm in Reed-Solomon decoders |

US7278086 | Jul 30, 2003 | Oct 2, 2007 | Hewlett-Packard Development Company, L.P. | Identifying uncorrectable codewords in a Reed-Solomon decoder for errors and erasures |

US7322004 | Apr 20, 2006 | Jan 22, 2008 | Marvell International, Ltd. | Efficient high-speed Reed-Solomon decoder |

US7353449 * | May 6, 2003 | Apr 1, 2008 | Thomson Licensing | Method of soft-decision decoding of Reed-Solomon codes |

US7539927 * | Apr 14, 2005 | May 26, 2009 | Industrial Technology Research Institute | High speed hardware implementation of modified Reed-Solomon decoder |

US7581155 | Dec 3, 2004 | Aug 25, 2009 | Electronics And Telecommunications Research Institute | Apparatus for FEC supporting transmission of variable-length frames in TDMA system and method of using the same |

US7805662 * | Feb 9, 2007 | Sep 28, 2010 | Sunplus Technology Co., Ltd. | Error correction code decoder |

US7870469 | Feb 12, 2007 | Jan 11, 2011 | Link—A—Media Devices Corporation | Parallel inversionless error and erasure processing |

US8201060 * | May 9, 2011 | Jun 12, 2012 | Ternarylocig LLC | Methods and systems for rapid error correction of Reed-Solomon codes |

US8365053 | May 27, 2009 | Jan 29, 2013 | International Business Machines Corporation | Encoding and decoding data using store and exclusive or operations |

US8365054 * | Nov 4, 2009 | Jan 29, 2013 | Lsi Corporation | Soft reed-solomon decoder based on error-and-erasure reed-solomon decoder |

US8433985 * | Mar 29, 2010 | Apr 30, 2013 | Intel Corporation | Error correction mechanisms for flash memories |

US8468434 * | Jan 21, 2011 | Jun 18, 2013 | Kabushiki Kaisha Toshiba | Error detection and correction system including a syndrome element calculating unit |

US8553968 | Feb 18, 2005 | Oct 8, 2013 | Western Digital Technologies, Inc. | Using optical character recognition augmented by an error correction code to detect serial numbers written on a wafer |

US8645803 | May 9, 2011 | Feb 4, 2014 | Ternarylogic Llc | Methods and systems for rapid error correction by forward and reverse determination of coding states |

US9032277 * | May 11, 2012 | May 12, 2015 | Altera Corporation | Parallel low and asymmetric rate Reed Solomon coding |

US20040255226 * | Feb 26, 2004 | Dec 16, 2004 | Keitarou Kondou | Decoder, error position polynomial calculation method, and program |

US20050149821 * | Dec 3, 2004 | Jul 7, 2005 | Hoon Lee | Apparatus for FEC supporting transmission of variable-length frames in TDMA system and method of using the same |

US20050166126 * | May 6, 2003 | Jul 28, 2005 | Stefan Muller | Method of soft-decision decoding of reed-solomon codes |

US20050188293 * | Feb 19, 2004 | Aug 25, 2005 | Quantum Corporation | Error correction decoder using cells with partial syndrome generation |

US20080082896 * | Aug 17, 2007 | Apr 3, 2008 | Broadcom Corporation | Burst error correction with offset for correction vector based on fire code |

US20100281344 * | Nov 4, 2010 | Gasanov Elyar E | Soft reed-solomon decoder based on error-and-erasure reed-solomon decoder | |

US20110202815 * | Aug 18, 2011 | Kabushiki Kaisha Toshiba | Error detection and correction system | |

US20110214038 * | Sep 1, 2011 | Ternarylogic Llc | Methods and Systems for Rapid Error Correction of Reed-Solomon Codes | |

US20110239094 * | Mar 29, 2010 | Sep 29, 2011 | Kwok Zion S | Error correction mechanisms for flash memories |

CN1653699B | May 6, 2003 | Jun 15, 2011 | 汤姆森特许公司 | Method of soft-decision decoding of reed-solomon codes |

EP1370003A1 * | Jun 7, 2002 | Dec 10, 2003 | Deutsche Thomson-Brandt Gmbh | Reed-Solomon Decoder |

EP1370005A2 * | Jun 4, 2003 | Dec 10, 2003 | Thomson Licensing S.A. | Reed-Solomon decoder |

WO2003096546A2 * | May 6, 2003 | Nov 20, 2003 | Marten Kabutz | A method of soft-decision decoding of reed-solomon codes |

Classifications

U.S. Classification | 714/784 |

International Classification | H03M13/15 |

Cooperative Classification | H03M13/1515, H03M13/154, H03M13/1555 |

European Classification | H03M13/15P4, H03M13/15P1, H03M13/15P7 |

Legal Events

Date | Code | Event | Description |
---|---|---|---|

Mar 23, 1999 | AS | Assignment | Owner name: STORAGE TECHNOLOGY CORPORATION, COLORADO Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:BOYER, KEITH G.;REEL/FRAME:009855/0597 Effective date: 19990322 |

Aug 12, 2005 | FPAY | Fee payment | Year of fee payment: 4 |

Jul 15, 2009 | FPAY | Fee payment | Year of fee payment: 8 |

Mar 13, 2013 | FPAY | Fee payment | Year of fee payment: 12 |

Rotate